Intelligent patent similarity searching method and device based on semantic retrieval

Abstract

The invention discloses an intelligent patent similarity searching method and device based on semantic retrieval. The method comprises the steps of obtaining patent text related fields of a target patent, performing word segmentation on the patent text related fields of the target patent, and performing synonym expansion on a word segmentation result of the patent text related fields of the target patent through a pre-trained synonym expansion model; according to the patent text related field word segmentation result and the synonym expansion result of the target patent, creating the Query of Elasticsearch, so that roughly recalled patent data is obtained based on Elasticsearch; calculating the cosine similarity between each patent text in the roughly recalled patent data and the target patent; according to the cosine similarity, sorting each patent text in the roughly recalled patent data, and returning a final sorting result; compared with the prior art, the retrieval efficiency of patent similarity search is higher, the recall accuracy of patent similarity search is higher, and the synonym expansion model can be continuously optimized according to incremental data.

Description

technical field [0001] The invention relates to the technical field of intelligent search, in particular to an intelligent patent similarity search method and device based on semantic retrieval. Background technique [0002] The essence of patent similarity search is an intelligent embodiment of patent retrieval, which is the process of selecting documents or information that meet a specific requirement from a large number of patent documents or patent databases according to one or several characteristics, and provides patent-based Ranked recall of document similarity size. [0003] Patent CN106372226B discloses an information retrieval device and method. The device includes: a receiving unit, which receives a specific patent number input by a user; a keyword acquisition unit, which automatically extracts keywords from the patent information corresponding to the specific patent number; a comparison unit, according to The degree of correlation between the keyword and the abo...

AI Technical Summary

Problems solved by technology

[0007] The present invention provides an intelligent patent similarity search method and device based on semantic retrieval to solve the problems of prior art in terms of patent similarity sorting accuracy, similar patent recall rate, and retrieval rate.

Images